S-5051-119
Read twice and referred to the Committee on Commerce, Science, and Transportation.
Sponsored by Mark Warner (D-VA)
What it does
The AI AGENT Act would require large online platforms (those with more than 50 million U.S. users per month) to maintain open, nondiscriminatory interfaces allowing users to authorize AI software agents — called "custodial user agents" — to manage their accounts, make purchases, post content, and handle other online interactions on their behalf. The bill would impose fiduciary-like duties on AI agent providers, require them to register with the Federal Trade Commission (FTC), and direct the FTC and NIST to establish technical standards and authentication procedures within 180 days to one year of enactment.
Who benefits
Individual users of large platforms who want AI agents to automate tasks like shopping, social media management, or financial account oversight. Startups and independent developers building AI agent products, who would gain guaranteed access to major platforms they might otherwise be blocked from. Consumers with disabilities who rely on automated tools to navigate online services. Small businesses that use AI agents to manage their online presence. Researchers and academics studying AI-platform interactions. Competing AI agent providers who would gain functional equivalence with any AI agents built by the platforms themselves.
Who is hurt
Large online platforms (e.g., major social media, e-commerce, and AI service companies) that would bear compliance costs and lose control over how third-party agents interact with their systems. Platform advertising businesses that depend on direct user engagement data, which AI agents may reduce or obscure. Platform-affiliated AI agent products that would face mandated functional equivalence with competitors. Users whose accounts could be exposed to new security or fraud risks if AI agents are compromised. Smaller platforms near the 50-million-user threshold that may face disproportionate compliance burdens. State governments whose tech-related privacy or interoperability laws could be partially preempted.
Supporters argue
Supporters argue that dominant platforms currently use technical barriers and restrictive terms of service to block third-party AI agents, locking users into their ecosystems and stifling competition in the emerging AI agent market. They contend that users should have the same right to delegate tasks to software agents as they do to human assistants, and that the bill's fiduciary-like duties — prohibiting agents from acting against user interests, selling user data, or sub-delegating authority without consent — provide robust consumer protections. They point to the EU's Digital Markets Act as evidence that interoperability mandates can increase competition without degrading platform quality.
Opponents argue
Opponents argue that forcing platforms to open their interfaces to registered third-party AI agents creates significant new attack surfaces for fraud, data theft, and manipulation that the bill's registration and certification regime may not adequately prevent. They contend that platforms' decisions about which agents may access their systems constitute editorial and operational discretion protected under Moody v. NetChoice (2024), and that mandating access amounts to compelled facilitation of third-party speech and conduct. They further argue that the FTC's broad rulemaking authority under this bill — covering authentication, technical standards, fee reasonableness, and agent registration — may face heightened judicial scrutiny under the major questions doctrine following Loper Bright v. Raimondo (2024).
Constitutional context
The Commerce Clause (Art. I, §8, cl. 3) provides the primary basis for federal regulation of large online platforms. However, Moody v. NetChoice (2024) established that platforms' content moderation and access decisions reflect protected editorial discretion, raising First Amendment questions about whether mandating third-party agent access constitutes compelled facilitation. Additionally, post-Loper Bright (2024), the FTC's broad delegated authority to set technical standards, fee structures, and registration terms will face independent judicial review rather than automatic deference.
Checks and balances
The FTC gains significant new rulemaking, registration, enforcement, and adjudicatory authority over both platforms and AI agent providers; checks include interagency coordination requirements, public notice obligations, a petition-and-appeal process for fee disputes, judicial review of FTC rules under the post-Loper Bright independent-judgment standard, and a partial preemption clause that preserves inconsistent state laws only to the extent they are consistent with the Act.
Historical precedent
The EU's Digital Markets Act (2022) imposed similar interoperability and access obligations on large "gatekeeper" platforms; no directly analogous U.S. federal statute mandating third-party AI agent access to online platforms has previously been enacted.